Merge with git+ssh://cvs.ucw.cz/projects/sherlock/GIT/sherlock.git#dev-make

[libucw.git] / lib / kmp.h

diff --git a/lib/kmp.h b/lib/kmp.h
index f4d14049235081ab90cf1c5123e87db1a2384d0d..8c8c525448f148f5b7ab9495415752e6736e8d86 100644 (file)
--- a/lib/kmp.h
+++ b/lib/kmp.h
@@ -16,12 +16,12 @@
  *
  *  This file contains only construction of the automaton. The search
  *  itself can be generated by inclusion of file lib/kmp-search.h.
- *  Separeted headers allows the user to define multiple search
+ *  Separeted headers allow the user to define multiple search
  *  routines for one common set of key strings.
  *
  *  Example:
  *
- *     #define KMP_PREFIX(x) GLUE_(kmp,x)
+ *     #define KMP_PREFIX(x) kmp_##x
  *     #define KMP_WANT_CLEANUP
  *     #define KMP_WANT_SEARCH // includes lib/kmp-search.h automatically
  *     #define KMPS_FOUND(kmp,src,s) printf("found\n")
@@ -51,14 +51,14 @@
  *    Basic parameters:
  *             KMP_PREFIX(x)           macro to add a name prefix (used on all global names
  *                             defined by the KMP generator); mandatory;
- *                             below we use P(x) alias
+ *                             we abbreviate this to P(x) below
  *
  *     KMP_CHAR                alphabet type, the default is u16
- *     
- *     KMP_SOURCE              user-defined text source; KMP_GET_CHAR must 
+ *
+ *     KMP_SOURCE              user-defined text source; KMP_GET_CHAR must
  *     KMP_GET_CHAR(kmp,src,c) return zero at the end or nonzero together with the next character in c otherwise;
  *                             if not defined, zero-terminated array of bytes is used as the input
- *     
+ *
  *     KMP_VARS                user-defined variables in 'struct P(struct)'
  *                             -- a structure describing the whole automaton;
  *                             these variables are stored in .u substructure to avoid collisions
@@ -131,15 +131,15 @@ typedef struct {} P(node_t);
 struct P(struct);
 
 struct P(state) {
-  struct P(state) *from;       /* state with previous character (forms a tree with null state in the root) */
+  struct P(state) *from;       /* state with the previous character (forms a tree with null state in the root) */
   struct P(state) *back;       /* backwards edge to the longest shorter state with same suffix */
-  struct P(state) *next;       /* longest shorter match (or NULL) */
+  struct P(state) *next;       /* the longest of shorter matches (or NULL) */
   P(len_t) len;                        /* state depth if it represents a key string, zero otherwise */
-  P(char_t) c;                 /* last character of represented string */
+  P(char_t) c;                 /* last character of the represented string */
   struct {
 #   ifdef KMP_STATE_VARS
     KMP_STATE_VARS
-#   endif    
+#   endif
   } u;                         /* user-defined data*/
 };
 
@@ -168,7 +168,7 @@ P(hash_hash) (struct P(hash_table) *t, struct P(state) *f, P(char_t) c)
 static inline uns
 P(hash_hash) (struct P(hash_table) *t UNUSED, struct P(state) *f, P(char_t) c)
 {
-  return (((uns)c) << 16) + (uns)(addr_int_t)f;
+  return (((uns)c) << 16) + (uns)(uintptr_t)f;
 }
 #endif
 
@@ -213,10 +213,10 @@ static inline void
 P(hash_init_key) (struct P(hash_table) *t UNUSED, struct P(state) *s, struct P(state) *f, P(char_t) c)
 {
   bzero(s, sizeof(*s));
-# ifdef KMP_INIT_STATE  
+# ifdef KMP_INIT_STATE
   struct P(struct) *kmp = (struct P(struct) *)t;
   { KMP_INIT_STATE(kmp, s); }
-# endif  
+# endif
   s->from = f;
   s->c = c;
   s->next = f->back; /* the pointers hold the link-list of sons... changed in build() */
@@ -224,7 +224,7 @@ P(hash_init_key) (struct P(hash_table) *t UNUSED, struct P(state) *s, struct P(s
 }
 
 #undef P
-#define HASH_PREFIX(x) KMP_PREFIX(GLUE(hash_,x))
+#define HASH_PREFIX(x) KMP_PREFIX(hash_##x)
 #define HASH_NODE struct KMP_PREFIX(state)
 #define HASH_KEY_COMPLEX(x) x from, x c
 #define HASH_KEY_DECL struct KMP_PREFIX(state) *from, KMP_PREFIX(char_t) c
@@ -256,7 +256,7 @@ struct P(struct) {
 #ifdef KMP_SOURCE
 typedef KMP_SOURCE P(source_t);
 #else
-typedef byte *P(source_t);
+typedef char *P(source_t);
 #endif
 
 #ifdef KMP_GET_CHAR
@@ -281,13 +281,13 @@ P(get_char) (struct P(struct) *kmp UNUSED, P(source_t) *src, P(char_t) *c)
 {
 # ifdef KMP_USE_UTF8
   uns cc;
-  *src = (byte *)utf8_get(*src, &cc);
+  *src = utf8_get(*src, &cc);
 # ifdef KMP_ONLYALPHA
   if (!cc) {}
   else if (!Ualpha(cc))
     cc = P(control)();
   else
-# endif  
+# endif
     {
 #     ifdef KMP_TOLOWER
       cc = Utolower(cc);
@@ -402,7 +402,7 @@ P(build) (struct P(struct) *kmp)
   null->back = NULL;
 # ifdef KMP_BUILD_STATE
   { KMP_BUILD_STATE(kmp, null); }
-# endif  
+# endif
   while (read != write)
     {
       struct P(state) *s = fifo[read++], *t;
@@ -425,7 +425,7 @@ P(build) (struct P(struct) *kmp)
        }
 #     ifdef KMP_BUILD_STATE
       { KMP_BUILD_STATE(kmp, s); }
-#     endif      
+#     endif
     }
 }